OSNews

I think this is great news as well, however, I wonder how well my apps will work with it. I know Adobe, due to their superior programming skills, does not work if you use a filesystem other than the default.

Why a graphics program concerns itself with the filesystem is entirely beyond me.

They probably mean "we" as in everyone involved wih ZFS somehow, which indeed are mostly Sun people.

AFAIK Noel Dellofano used to work for sun microsystems so she might very well be one of the people that created ZFS.

A simple search on Noel Dellofano finds his blog:
http://blogs.sun.com/dellofano/

Notice the address. When he says "we" he simply means "we the developers of ZFS", not company X or Y.

Some points:-

- His last post was in 2005 before anyone else had picked up ZFS.

- His blog is on the sun.com domain

- Obviously he is referring to Sun developers.

It's simple. He's one of the creators on the ZFS project who then took a job at Apple. The WE work both at SUN and Apple.

I agree its simple ZFS was created by Sun employees. Sun paid his salary, held the rights to the software and released it to the world for all to benefit. The fact that he moved to another company has nothing to do with the fact that Sun blew "away 20 years of obsolete assumptions" and not Apple and it certainly should not be represented as "We" on an Apple/Mac website as this clearly gives the wrong impression.

If you read the page where the quote in the article is actually taken from, you'll notice this note:

**Note the above description is taken directly from the ZFS Open Solaris site What is ZFS page

http://trac.macosforge.org/projects/zfs/wiki/whatis

Edited 2008-01-11 23:06 UTC

Thank you I am really enjoying these videos.

A Mac mini doesn't fit for ZFS. It's for really big servers.

ZFS is for desktops, laptops, small servers, medium servers and really really big servers. There is nothing about ZFS that makes it un-useable on a mac-mini. Sure there are features in ZFS which only really come into their own on large servers. But there are also many features that make sense on all computers. The storage pools that the original poster talked about is a perfect example.

Which is great for Sun, who actually exists in datacenters and does deserve a little bit of benefit for creating it in the first place.



Yes it will, as is most technology that they assimilate, but it will not change anything. Seriously, 99.9% of the people cheering about this for Apple sadly have no actual idea of what it does differently, or how it will make a real difference to them.

It shouldn't really matter whether Apple users understand ZFS vs xFS. Indeed, Apple sometimes assimilates certain technologies, improves them and make them available on OS X. And by improving, Apple usually means integrating the technology and making it accessible to all or most users.

One perfect example is Time Machine, which essentially is just hardlinks + rsync. So no, Time Machine is hardly revolutionary; five years ago I wrote a simple 50-line bash script that does exactly the same. The real improvement of Time Machine is its integration into the OS and its accessibility toward non-tech users. That is Apple's real surplus; not only with Time Machine but also with many other technologies like ZFS (if done well). That's why it shouldn't matter whether users understand the technical merits of ZFS; ideally it should become obvious during day-to-day usage.

I'm an Apple and Linux user who knows what ZFS is capable of.

I've had to deal with stupid #$%^ RAIDs and tape backups inside a data center.

It will not alleviate all issues, such as hardware failure, but it will get many issues, like silent data corruption, off our backs.

The self-healing model in-and-of-itself will be worth it to businesses small, medium and large. In many ways the operating system is becoming irrelevant. The data we produce, manipulate, store and retrieve is more important.

Apple was just parroting Sun's material. If Apple can integrate Sun's open technology into its own and make it work then it is good for those who use ZDF.

Indeed. As a Linux user, it is frustrating to me that that solution in search of a problem, reiser4, has gotten so much hype and attention, and yet we are nowhere *near* having a filesystem like ZFS which actually solves problems the we face everyday, and which will only be increasing in importance as time goes on.

Unfortunately, there are multiple layers of difficulty. The licensing, of course. And then the patent issues. And if, by some miracle, those two things got resolved, we would still be facing the issue of ZFS being, as Andrew Morton put it, "a rampant layering violation".

Ext4 looks nice, but does not really address all the issues. Our best bet for the future looks, to me, like Oracle's/Chris Mason's btrfs.

Chris has been known to do quality work, and make rapid progress, so maybe with Oracle behind it btrfs will be ready sometime in the not *too* terribly distant future.

Edited 2008-01-12 17:26 UTC

Replying to myself... in case anyone is interested, Chris actually has a proposed development timeline available:

http://tinyurl.com/3dod68

SBERGMAN27

"Unfortunately, there are multiple layers of difficulty. The licensing, of course. And then the patent issues. And if, by some miracle, those two things got resolved, we would still be facing the issue of ZFS being, as Andrew Morton put it, "a rampant layering violation"

About the rampant layering violation, Jeff Bonwick at Sun bloggs about it. In short, he says that Morton hasnt really understand ZFS, and therefore his claim is not really correct:
http://blogs.sun.com/bonwick/entry/rampant_layering_violation

Fair enough, but kind of my point. You talk about datacenters. Which organization is more likely to wind up in a datacenter, Sun or Apple? Seriously. Are your bosses going to say, "hey, let's move to Macintosh hardware for our mission critical operations now that they have ZFS!" ?



No, it's not. It's good for Apple users who want to use ZFS. Apple's adoption of ZFS holds no benefit for Sun's users of ZFS, any more than Apple's adoption of BSD or GNU technologies benefits those organizations. Apple won't be signing any coprighted code over to Sun, so there's very little chance that any modifications or changes Apple makes to their ZFS implementation will benefit Sun, though maybe the community openSolaris users.

At any rate, the point I was making was that many people are jumping cartwheels as if Apple has broken new ground, without really understanding what they're tubthumping. I've got no issue with Apple adopting ZFS, I don't think it's a bad thing, but it means zilch to the average user. A bunch of power users frequenting this forum and espousing the advantages of ZFS doesn't really change the fact that it still means zilch to the average user. It's like bragging about adopting dtrace; it's a great and powerful feature, but means zilch to the average user.

Simply trying to put some perspective on the hype, that's all.

but means zilch to the average user.

An average user probably won't know the difference between ZFS and HTML, but, it will still benefit them. ZFS does still protect them against silent data corruption and all that, and if Apple (or any other company/Linux distro/etc) can make it really simple to just add a new hard drive to the pool the average users will benefit from that too.

ZFS is definitely a work in progress as you can see from threads at the opensolaris help and opensolaris zfs mailing lists.

But it is fine for desktops and it doesn't NEED all that memory, and it is pretty stable.

It's crap on the desktop and need lots of gigabytes to run almost stable (just for the filesystem!). And even then it is not! It's even in Solaris a work in process and far away from any reliability

You do realize that if a filesystem is a work in progress it means they are trying to improve it even more, like f.ex. optimize it some more? Reiser3 has been a work in progress for years and it is VERY stable anyway. And why should ZFS be "crap on the desktop"? You don't clearly understand the benefits of pooled storage for regular users. And no, it doesn't take gigabytes of memory. The OS itself could, but not the filesystem layer.

Silent data corruption can only be healed if there's redundancy. That requires either ditto blocks for data or actual physical redundancy. Latter is unlikely to be seen in generic consumer machines, former probably just on certain filesystems created for that purpose (e.g. the document folder).

ZFS stores copies of written blocks in unused blocks and as such resolves the problem. There are of course CRC checks and such also. I remember having read that it stores actually three copies of a single block all spread out in the pool but those extraneous copies aren't counted as used disk space and so can be overwritten by data if space on the pool is getting small.